Text classification on the Instagram caption using support vector machine

Abstract

Instagram is Top 10 the most popular social networks worldwide with over 1 billion monthly active users. Instagram is especially prevalent in the United States, India, and Brazil, which have over 130 million, 100, and 91 million Instagram users each. And the fact that so many people use Instagram, Social Media Marketing is also one of the goals of people using social media as a place to market their products. To help people to find out what's trending on Instagram, this paper using text classification to categorizing Instagram caption into organized groups (fashion, food & beverage, technology, health & beauty, lifestyle & travel). In this paper, we are testing the Support Vector Machine algorithm to classify the trending on Instagram using 66171 captions. The study used the TFIDF method (Term Frequency times Inverse Document Frequency) measure and used variations of percentage for splitting data. The results showed that the use of SVM with a percentage ratio of 70:30 resulted in higher accuracy compared to others.